When AI Becomes Undetectable: The Trust Crisis No One Is Ready For

What happens when we can no longer tell what is real, and what does that mean for the institutions we lead?
TL;DR
- We are approaching a moment where distinguishing between human-created and AI-generated content will become functionally impossible. We are not there yet. But we are close enough that institutions not preparing now will be entirely unprepared when it arrives.
- Detection is not a strategy. The tools that promise to identify AI-generated content are already failing. The institutions betting on detection are betting on a losing hand.
- Trust is the new infrastructure. When you can't verify content by looking at it, you verify it by knowing the source. That shift changes everything about how institutions operate.
- Higher education must move from detection to verification. Build systems where trust is embedded in process, not dependent on tools that can't keep up.
- This is not a post about panic. It is a post about clarity. The leaders who understand what is actually happening will be the ones who can guide their communities through it.
There is a moment in every Where's Waldo book where the longer you look, the less certain you become. You start to question things you were sure about a moment ago. A red stripe becomes ambiguous. A face you thought was Waldo turns out to be someone else entirely.
That is where we are with AI right now.
Not the fun kind of uncertainty. The kind that has real consequences for how institutions function, how students learn, and how leaders make decisions.
We are approaching a moment, faster than most people realize, where distinguishing between human-created and AI-generated content will become functionally impossible. We are not there yet. But we are close enough that the institutions not preparing for it now will find themselves entirely unprepared when it arrives.
This is not a post about panic. It is a post about clarity. Because the leaders who understand what is actually happening will be the ones who can guide their communities through it.
UNDETECTABLE AI IS ALMOST HERE
What Is Undetectable AI and Why Does It Matter Now
[]()
Two terms come up repeatedly in this conversation and they are often used interchangeably. They are not the same thing.
- Undetectable AI refers to written content generated by an algorithm designed to imitate human writing so closely that no tool, reader, or detection system can reliably identify it as machine-generated. We are already seeing early versions of this. The detection tools available today are inconsistent, frequently wrong, and increasingly outpaced by the generation tools they are trying to catch.
- Deepfakes refer to AI-generated video and images that portray real people doing or saying things they never did or said. The technology has advanced from obviously artificial to genuinely indistinguishable from real footage in a matter of a few years.
In both cases, the intent is the same: to make the audience believe something that is not true.
WHAT THIS TELLS LEADERS
What makes this moment different from previous media literacy challenges is the scale and the speed. A single person with a free AI tool and a smartphone can now produce content that would have required an entire production team five years ago. And they can distribute it to millions of people before any fact-checker has had time to open their laptop.
For higher education, this is not an abstract future problem. It is arriving in your classrooms, admissions offices, student affairs offices, and faculty meetings right now.
SEEING IS POWERFUL EVEN WHEN WE KNOW IT IS NOT REAL
We Are Already Losing the Ability to Tell What Is Real
The Guardian has described the effort to distinguish real from AI-generated images as a race against time. That framing is important. It is not a race we are winning.
What makes this particularly difficult is not just the technology itself. It is human psychology.
Research on deepfakes has revealed something deeply uncomfortable: showing people a warning that a video is fake before they watch it does not reliably change how they process it. The social and emotional experience of watching the video overrides the rational knowledge that it is fabricated. People will still reference it as fact even after being told it is not.
Henry Parker, head of government affairs at a leading fact-checking organization, put it plainly: if you tell someone they are looking at a deepfake before they watch it, the psychological experience of watching that video is still powerful enough that they will treat it as real. The only practical question becomes how quickly you can remove the content from circulation.
That is a remarkable and sobering statement. It means the solution is not simply better labeling or more warnings. The problem runs deeper than that. It is rooted in how human beings process information — emotionally first, rationally second.
[]()
WHAT THIS TELLS LEADERS
For higher education leaders, this raises an urgent question: what are you teaching your students about how their own minds work alongside media? And what are you doing to build institutional resilience before a deepfake of your president, your faculty, or your campus lands in your community's feeds?
Is your institution prepared for AI or just reacting to it?
Most leaders we talk to say the same thing: "We know we need a strategy. We just don't know where to start."
You do not need to have all the answers today. You need a thought partner who has helped institutions like yours find their footing.
[Let's talk about where your institution stands →](/contact)
Here is what getting started actually looks like:
- A clear picture of where AI is already active in your institution
- A framework for making decisions before a crisis forces your hand
- Faculty, staff, and leadership who can navigate AI with confidence and judgment
- Policies that protect your community without shutting down innovation
AI FLUENCY IS A SAFETY ISSUE
The Trust Crisis Did Not Start With AI
[]()
To understand where we are, it helps to understand how we got here.
For decades, journalism operated as a trusted civic institution. In 1972, CBS anchorman Walter Cronkite was voted the most trusted person in America. Presidents referenced him as a steady voice in uncertain times. That level of institutional trust in a media figure is almost impossible to imagine today.
What changed was not one thing. It was a cascade.
The arrival of 24-hour cable news created what journalists call "feeding the beast" — the relentless demand for content around the clock. Before that, newspapers published once or twice a day. Broadcast networks aired news at fixed hours. That rhythm gave journalists time to verify, to fact-check, to correct before publishing. The pressure was to get it right by deadline, not to get it first by the minute.
When the cycle became continuous, speculation filled the gaps. Editorial opinion blurred into news coverage. Stories that would not have merited significant attention under the old model got elevated because there was airtime to fill. Trust began to erode.
Then came social media, which removed the institutional filter entirely. Anyone could publish anything and reach millions of people instantly. The architecture of social platforms rewarded engagement over accuracy. Outrage traveled faster than correction.
PAY ATTENTION TO THE AMBIENT SHIFT
What Happens to Higher Education When Reality Becomes Negotiable
Higher education has always operated on a set of assumptions about shared reality. Evidence-based inquiry. Peer review. The idea that facts can be established and built upon. The entire academic enterprise depends on a baseline agreement that some things are true and some things are not.
When that baseline erodes — when students, faculty, and community members genuinely cannot agree on what is real — the consequences reach into every corner of institutional life.
Consider a few scenarios that are not hypothetical:
A deepfake video of a university president making a statement they never made circulates on student social media before anyone can respond. Even after the university issues a correction, a significant portion of the campus community continues to reference the video as evidence of what leadership "really thinks."
An AI-generated essay submitted by a student is flagged by a detection tool. But the tool has a documented error rate of 15 percent. The faculty member now faces an impossible choice: accuse a student of academic dishonesty based on a tool that could be wrong, or ignore the flag and potentially reward dishonesty.
A prospective student receives what appears to be an official email from your admissions office offering a scholarship that does not exist. The email was generated by AI and designed to harvest personal information.
None of these scenarios require sophisticated actors. They require only access to tools that are freely available right now.
The question for leaders is not whether these things will happen. It is whether your institution has a framework for responding when they do.
FORMATION OVER INFORMATION
This Is Not a Technology Problem. It Is a Formation Problem.
Here is the part where you might expect a list of detection tools and software solutions.
That is not what this post is going to give you.
Not because tools do not matter. They do. But because if your institution's response to undetectable AI is purely technical, you will always be behind. The generation tools will always outpace the detection tools. That is the nature of the race.
The more durable response is formation.
Formation means building in your students, your faculty, and your staff the habits of mind that allow them to navigate uncertainty with judgment rather than panic. It means teaching people not just how to use AI tools but how to think alongside them. It means building communities where critical evaluation is a shared expectation, not an individual burden.
[]()
WHAT THIS TELLS LEADERS
This is not new work. It is the oldest work of higher education. What is new is the urgency.
We are moving toward a moment where societal paranoia — the condition where people cannot agree on basic factual information and approach every source with suspicion — becomes the default mode of public life. That paranoia is already visible on campuses. Students who distrust institutional communications. Faculty who question the authenticity of student work. Leaders who cannot get traction on important decisions because every message is immediately contested.
Formation is the antidote to paranoia. Not information — formation. The difference matters.
Information tells people what to think. Formation develops people who know how to think. That is the work higher education was built for. And it has never been more urgent than it is right now.
Your institution doesn't need one AI expert; you need a network of AI First Responders.
Stop treating AI adoption like a project. Start building a culture of confidence.
AI First Responder is the credential that changes that.
It's not about IT. It's about people. It's about equipping faculty, staff, and leadership to respond with clarity, confidence, and integrity when colleagues need guidance. This is the work that happens in offices, in hallways, during advising appointments, and at faculty meetings.
[Learn about the AI First Responder program →](/ai-first-responder)
AI First Responders do:
- Normalize confusion instead of escalating fear
- Help colleagues think with AI rather than hide it
- Model ethical, transparent AI use
- Know when to reassure and when to escalate
They don't:
- Police or interrogate colleagues
- Make assumptions about intent
- Make colleagues feel ashamed for asking
- Override institutional guidelines
This isn't a one-time workshop. It's a structured, practice-based credential built for the real work of institutional leadership.
What Higher Ed Leaders Can Do Right Now
You cannot fix this alone. No single institution can fix this alone. But you can build a more resilient community, and you can start today.
Audit your institutional communications for vulnerability. Where could a deepfake or AI-generated message cause the most damage in your community? Admissions communications. Presidential statements. Emergency alerts. Financial aid correspondence. Know your vulnerabilities before someone exploits them.
Build AI literacy into the curriculum, not as a standalone course but as a thread. Every discipline touches questions of evidence, source evaluation, and truth claims. AI literacy belongs in those conversations across the curriculum, not siloed in a technology elective.
Train faculty and staff on detection tools with honest conversations about their limits. Detection tools exist and have value. But faculty who believe those tools are reliable and infallible will make consequential errors in academic integrity cases. Honest professional development matters more than false confidence.
Develop institutional response protocols for deepfakes and AI disinformation. You need a plan before you need it. Who is the decision-maker when a deepfake circulates? What is your communication strategy? How do you balance speed with accuracy? Work this out in advance.
Invest in the formation of critical thinkers, not just critical thinking skills. Skills can be taught in a workshop. Formation takes time, community, and repetition. It requires institutions to model the habits they want students to develop. That starts with leadership.
If your institution is navigating the AI trust crisis and you are not sure where to start, this is exactly the work I do with higher education leaders.
[Let's chat](/contact)
COMMON QUESTIONS
Frequently Asked Questions
What is undetectable AI and how is it different from deepfakes?
Undetectable AI refers to written content generated by algorithms designed to mimic human writing so closely that readers and detection tools cannot reliably identify it as machine-generated. Deepfakes refer to AI-generated video and images that portray real people in situations that never happened. Both are designed to deceive, but they operate in different media and present different challenges for institutions trying to respond to them.
Can AI detection tools reliably identify AI-generated student work?
Not reliably, no. Current AI detection tools have meaningful error rates and are frequently outpaced by advances in generation technology. This does not mean institutions should ignore detection tools entirely, but it does mean that academic integrity policies built entirely around detection tool results are on unstable ground. The more durable approach combines tool use with faculty training, assignment design that makes AI generation less useful, and honest conversations with students about academic integrity.
Why is this described as a formation problem rather than a technology problem?
Because technology solutions — detection tools, labeling systems, platform policies — will always lag behind the generation tools they are trying to address. The more durable institutional response is building communities of people who have the habits of mind to navigate uncertainty with judgment. That is the work of formation, which has always been at the heart of higher education's mission. The urgency is new. The work is not.
How should higher education institutions respond when a deepfake circulates on campus?
Speed and accuracy are both important, but accuracy takes priority. A rapid response that is wrong creates more damage than a slightly slower response that is right. Institutions should develop deepfake response protocols in advance, designating decision-makers, communication channels, and verification processes. The response should include a clear, direct statement from leadership, distribution through every trusted institutional channel simultaneously, and a follow-up communication that explains what happened and what the institution is doing to prevent recurrence.

Let's talk.
If you are leading an institution through this moment and you are not sure where to start, that is exactly what I do.
[Request a consultation](/book-a-call)
Sarah Gibson is Director of the School of Communication and Faculty Fellow for AI at Lipscomb University and founder of PivotPoint EDU. She helps higher education institutions and nonprofits move from AI anxiety to AI strategy with clarity and confidence. Learn more.
Comment 0 Comments JD